Sept. 30, 2024: $AERO-drama π¬πΏ
A tale of LVR, Toxic Flow, and next generation AMM architecture
DISCLAIMER: author has exposure to mentioned tokens CRV, VELO + AERO (but not UNI)
Although nobody really wins in π beefs, this weekendβs flare-up was worth highlighting for the reason that it led to a good discussion about the theoretical intricacies of the bleeding edge of AMM architecture.
The opening salvo was fired by former Uniswap researcher Austin Adams, who has since left to catpost and build independent AMM design company Whetstone. Read the thread in its entirety to understand the context, which weβll break down in greater depth toward the end of the article after we review the flame wars.
The *drome community is naturally protective of their protocolβs reputation, just as one would hope and expect to see (would anybody trust their funds to a protocol that simply folded on confrontation?)
No surprise the hostilities flared up as if somebody had tossed a can of gasoline into a BBQ pit. Below weβve done our best to snapshot all the key skirmishes, without adding too much of our leftwit commentary. Begin warming your popcorn oilβ¦
The Aero community was quick to weigh inβ¦
Counter-threadsβ¦
Comment threadsβ¦
Dashboardsβ¦
Feeposting:
And weβre close to the end of the hostilitiesβ¦
Not yet, just a bit furtherβ¦
OK, letβs stop here. The βDynamic Fee Modelβ has been invokedβ¦
Dynamic Fee Model
So what is wrong with Dynamic Fees? Weβve seen dynamic fees leading to great APYs on Curve lately, is this a bad thing?
Curve Founder Michael Egorov enters the chatβ¦
The answer is that dynamic fees are in fact good, but in terms of the frontiers of DEX design itβs already yesterdayβs newsβ¦
You can see how, the conversation starts to turn productive around the time it turns to DEX architecture.
In truth, itβs an important subject as few understand the nuances.
Admittedly, weβre not among this elite cadre of under a dozen right-curvers. We promise that as soon as we fully understand this crypto ish, weβll shut down our newsletter as we retire to get rich running anonymous MEV bots in the background. But weβre not there yet, and as long as we have more to learn, weβll keep churning out paragraphs for your benefit.
Now what about this LVR term that got thrown about.
LVR
Maybe you typed βLVRβ into Google and checked out of the conversation when you learned it referred to an American footie club.
In DeFi, LVR stands for βloss-versus-rebalancing.β Last year it was memorably derided as an impossibly theoretical construct:
Despite being imperfect, it persists because it works pretty well when AMM engineers optimize towards this concept. Itβs roughly analogous to how the discipline of economics draws its best results when it operates on the premise that microeconomic actors are mostly rational, even though this premise is clearly countermanded by reality.
Once again Curve inserted itself into this debate, noting that its optimizations have historically been to maximize returns given only the presence arbitrage trading.
Which prompted some comments.
And user lmy1352 added productively to the conversationβ¦ π
For the extreme left-curve LPs (like the author, who makes like one transaction each month if gas fees dip,) it would be effectively impossible to approach this theoretical optimum. The passive LP generally does not care about βtoxic flow.β
In the universe where multiple types of AMMs coexist, LPs can theoretically bounce among different types of pools depending on expected volatility:
Which brings it all back to the question underlying the original premiseβ¦ is it possible for βvolumeβ to be bad?
Finally, what is being done at the AMM architecture level to solve the problem? Ser Adams takes the opportunity to point to the amAMM, an MEV capturing AMM:
The result will be an emergence of newer DEX structures, for which the metric for success is less likely to be raw volume.
Disclaimers! We did our best to capture everything here, but please drop clarifying comments below!